Notes: Session 10: Governance of data analytics
2026-05-12: 85 min
| Time (min) | Duration | Topic | Additional materials |
|---|---|---|---|
| 0–5 | 5 | Introduction | |
| 5–20 | 15 | Industry-level perspectives | Müller et al. 2018; Oesterreich et al. 2022 |
| 20–50 | 30 | Organizational perspectives | Kunz et al. 2025; TDWI Analytics Maturity Model |
| 50–80 | 30 | Responsible deployment and explainability | NIST AI RMF; GDPR; EU AI Act |
| 80–90 | 10 | People perspective | BCG AI Radar 2025; outlook discussion |
How machine learning changes value creation
What would you expect, how could ML affect the value creation process?
People
TODO: Last slide: discussion prompt on the evolving role of people in analytics.
Exercise
2026-05-12: 75 min (with two groups)
LO: contextualize the use of analytics in organizations, with respect to the organizational strategy, the value creation process, the specific analytical questions, …
TODO: include examples like ride hailing (Uber), Airbnb, or marketplaces like Amazon/ebay
TODO: add instructions on how to share results (e-mail/teams)
KPI/alignment framework: The strategy map approach seems to be a relatively good fit: Peppard, J., Ward, J., & Daniel, E. (2007). Managing the realization of business benefits from IT investments. MIS Quarterly Executive, 6(1), 1-11. - https://oro.open.ac.uk/11227/2/Peppard-Ward-Daniel_BMc_edits_JW230307.pdf https://en.wikipedia.org/wiki/Strategy_map
Provide specific materials as a starting point, such as the Medium engineering blogs - such as Netflix, LinkedIn, (select: data analytics and firm specific)
Examples:
- Netflix: Thumbnail personalization
- Airbnb.tech
- Uber engineering
- Spotify engineering
- LinkedIn engineering blog
- Engineering at Meta
- DoorDash engineering
- Shopify engineering blog
Note: removed Capital One from the list of examples (the example would focus primarily on lending decisions, which were already covered before in the lecture)
Additional materials
Ethics/Fairness: https://developers.google.com/machine-learning/crash-course/fairness
https://github.com/cs124/labs/blob/main/Lab2_LogisticRegression_Solutions.md#ethics-discussion
Deployment -> AI for generating and understanding analytical code. https://www.datacamp.com/de/tutorial/one-hot-encoding-python-tutorial - see “AI explain code” https://www.datacamp.com/datalab https://www.datacamp.com/datalab/ai
-> connect to AGI debates (jagged frontier…), which unlocks generation, but places new burdens on validation and testing
Reliability: monitoring/ensuring that the data input remains stable/in-line with the model (e.g., one-hot encoded variables with the specific levels for regression)
Fairness: https://developers.google.com/machine-learning/crash-course/fairness https://rodneybrooks.com/rodney-brooks-three-laws-of-artificial-intelligence